Problem with faked-system-time option

[Jerome Baum]

>> Suppose the party who originated the document to be signed
>> subsequently presents (possibly faked) evidence showing the document
>> to have been prepared later than the signature timestamp. The signer
>> is now unexpectedly in the position of having to prove something.
>
> First: That is no contradiction to what I have said. Have a look at the
> offline world: You never(?) sign anything in order to be able to prove that
> you have done or have to do something. You sign in order for others to be able
> tp prove that you have done or have to do something.

Addressing your "?", you might sign a memo regarding a phone call.
Three years later in court, nobody will believe that you can recall
exactly what you and the other party said. However, a written note,
bearing your signature and claimed date, makes your statement that
much more believable. I think we should all remember that proving
something is, like security, not a boolean. Your signature on the memo
isn't very strong as proof. It's also not worthless!

> Second: I really doubt that your case is a practical problem. As I said: The
> other one's interest is usually to be able to prove that you have signed and
> not that you haven't.

+1

> And even if you were "accused" of having signed with a faked system time: So
> what? This accusation is very dangerous, BTW. Everyone can easily get
> trustworthy timestamps for his documents or signatures. So you present a
> "proof" that the other one has manipulated and he has a better proof that your
> "proof" is fake? Faking such a proof is probably much worse than faking a
> timestamp for a normal signature.

Usually it'll be something like "false accusation", "falsification of
documents", etc. You can't say absolutely (without context) that the
fake proof is "much worse" than a fake timestamp. It really, really
depends on the context. Consider that the fake timestamp could also be
considered falsification of documents. An excellent source is the
German Criminal Code, section 267 ("Forgery"), and English translation
of which can be found at:
http://bundesrecht.juris.de/englisch_stgb/englisch_stgb.html#StGBengl_000P267

By the way, is there some Internet law mailing list around? I'm happy
with the "off-topic but Internet law related" posts but we might as
well cross-post for even more insight.

> An idea: I suggest a standardized signature notation like "timestamp". It
> would indicate that you don't make any statement about the signed content
> (which even may be encrypted, even against you) but just confirm the time of
> existence. That would solve (or reduced) the recently mentioned problem "You
> don't know what you sign".

Why modify the standard? Look at stamper (itconsult.co.uk), which just
adds some text to the signed content about no warranty etc. Should
suffice. Of course, not easy to parse, so obviously limited mostly to
human interpretation.

> The real problem is IMHO that keys can be revoked (without any bad intention).
> If you don't have a third party timestamp or something similar to prove that
> the signature has been made before the key was revoked then the signature is
> nearly worthless.

Yes! Says also German legal code when it comes to electronic
signatures. You're supposed to get timestamps from a third-party, and
regularly renew those timestamps. Not just for key revocation,
consider algorithm "decay" and the implicit invalidity introduced by
that.

-- 
Jerome Baum
tel +49-1578-8434336
email jerome at jeromebaum.com
web www.jeromebaum.com
--
PGP: A0E4 B2D4 94E6 20EE 85BA E45B 63E4 2BD8 C58C 753A
PGP: 2C23 EBFF DF1A 840D 2351 F5F5 F25B A03F 2152 36DA